A more profound issue is whether a financial system based on narrow banking could allocate capital efficiently.

Here there are two opposing risks. The first is that the supply of funds to riskier, long-term activities would be greatly reduced if we did adopt narrow banking. Against this, one might argue that, with public sector debt used to back the liabilities of narrow banks, investors would be forced to find other such assets.

The opposite (and greater) risk is that the fragility of banking would be re-invented, via "quasi-banks". This is what has just happened, after all, with "shadow banking". In the end, those entities, too, have been rescued. The big point is that a financial structure characterised by short-term and relatively risk-free liabilities and longer-term and riskier assets is highly profitable, until it collapses, as it is rather likely to do.

The answer to the second dilemma is to make banking illegal. That is to say, financial intermediaries, other than narrow banks, would have the value of their liabilities dependent on the value of their assets. Where assets could not be valued, there would be matching lock-up periods for liabilities. The great game of short-term borrowing, used to purchase longer-term and risky assets, on wafer-thin equity, would be ruled out. The equity risk would be borne by the funds' investors. Trading entities would exist. But they would need equity funding.